🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Successfully defended my thesis today! Extremely grateful to my amazing advisor, Kamalika Chaudhuri, whose unwavering support, insights, teachings and motherly love made all this possible.. Even the darkest of days seemed lighter & hopeful after spending just 5 mins with her! It has been

Congrats Chhavi Yadav

How can we use explainability methods when we can’t trust the explainer? Check out ExpProof paper by Chhavi Yadav Kamalika Chaudhuri Dan Boneh Evan Monroe Laufer

Upon graduation, I paused to reflect on what my PhD had truly taught me. Was it just how to write papers, respond to brutal reviewer comments, and survive without much sleep? Or did it leave a deeper imprint on me — beyond the metrics and milestones? Turns out, it did. A

Nice PhD retrospective by my (now former :-)) PhD student Chhavi Yadav :

As a PhD student, you don’t have to be the fastest, the most productive, or even the most charismatic- there are no limits to these things. However, it makes me happy when I see a PhD student to be introspective.

Very nice and thoughtful post by Chhavi. Now also a guest post on my blog: kamathematics.wordpress.com/2025/06/08/gue…

Got a shiny parting gift from UCSD CSE, the 2025 Doctoral Award for Excellence in Research! 🎓✨ Grateful for the recognition of my work on Incentive-Aware Trustworthy AI :) Awards certainly weren’t the goal… but hey, I’ll take the trophy and run! 😄

Proud advisor moment: a big congrats to my student Chhavi Yadav for the 2025 UCSD CSE Doctoral Award for Excellence in Research!

Commencement! 🎓 Draped a saree in the loving memory of my grandparents who would’ve been overjoyed to be here today ❤️

Distribution Inference for LLMs! During my PhD, I constantly got asked 'but what about LLMs?' — thrilled to see recent work tackle this 🔥 Pengrun Huang Chhavi Yadav Ruihan Wu and Kamalika Chaudhuri explore leakage of distributional properties from fine-tuning data

I went to take a passport photo a few days ago. A lady clicks the photo and hands over to me. I look at it and tell her the picture is blurry — anyone could tell looking at the picture. She says without even looking at it “the AI says it’s fine so it is fine” 😆🥲🤷🏻‍♀️

if you find this area exciting, there is some fascinating concurrent work which examines the problem of inferring training *distributions* from LLMs: x.com/iamgroot42/sta… (cc Chhavi Yadav Kamalika Chaudhuri et al!)